MiniCPM5-2B-heretic

A decensored variant of openbmb/MiniCPM5-2B, produced with Heretic v1.4.0 (directional ablation / "abliteration", per-layer refusal directions). Refusal behavior is suppressed via targeted weight edits to the attention output and MLP down-projections rather than fine-tuning, so the base model's knowledge and instruction-following are left largely intact.

Who this is for: developers who want MiniCPM5-2B's 2B-class capabilities — agentic tool use, code generation, long-context, and hybrid Think / No-Think reasoning — without the refusal guardrails. Ideal for local agents, roleplay, research on alignment/refusal mechanics, or any use case blocked by over-refusal. Runs comfortably on consumer GPUs and small machines via GGUF.

Runs on your gaming PC

Full GGUF ladder included — pick the quant that fits your card:

Weights only, at this model's 2B native size; add ~1 GB for context. OOM? Drop one quant level. Headroom to spare? Go one up.

Why abliteration instead of fine-tuning

Fine-tuning a "helpful" persona on top of RLHF'd refusals fights the base model's training and tends to degrade coherence. Abliteration instead finds and edits the specific weight directions responsible for refusal, leaving the rest of the network (and its capabilities) untouched. See the Heretic repo and the original abliteration writeup for the mechanism.

Abliteration parameters

Performance

Refusals dropped from 99 to 4 out of 100 while preserving MiniCPM5-2B's tool-use, code, and reasoning abilities.

Made with ❤️ by RACER IS OP (saidutta69) — follow for more uncensored models

Files

GGUF quantizations

Full quantization set (14 quants + F16) produced with llama.cpp.

MiniCPM5 architecture — loads natively in llama.cpp / Ollama / LM Studio / Jan.

Run llama serve -hf saidutta69/MiniCPM5-2B-heretic to pull the default quant.

Quickstart

# llama.cpp
llama serve -hf saidutta69/MiniCPM5-2B-heretic

# transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "saidutta69/MiniCPM5-2B-heretic"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="auto",
    device_map="auto",
)

messages = [{"role": "user", "content": "Who are you? Please briefly introduce yourself."}]
inputs = tokenizer.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    enable_thinking=False,   # set True for Think mode
    return_dict=True,
    return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=128)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

Also runnable via Ollama, LM Studio, Jan, vLLM, SGLang. For tool/function calling, SGLang is the recommended backend; MiniCPM5 emits XML-style tool calls that SGLang's built-in minicpm5 parser converts to OpenAI-compatible tool_calls.

FAQ

What is MiniCPM5-2B-heretic?

MiniCPM5-2B-heretic is a decensored build of openbmb/MiniCPM5-2B produced with Heretic 1.4.0 (directional ablation). It is the same weights with the refusal direction removed, not a new fine-tune, so its capabilities track the base model.

Is MiniCPM5-2B-heretic actually uncensored?

Yes. On Heretic's 100-prompt refusal harness it drops from 99/100 to 4/100 at KL divergence 0.0646 from the base model. These are measured values, not a claim; reproduction data is in reproduce/.

How do I run MiniCPM5-2B-heretic locally?

llama serve -hf saidutta69/MiniCPM5-2B-heretic:Q4_K_M pulls the default quant; the repo ships 15 GGUF quants (see Files).

Responsible use

Refusal suppression is deliberate and works as intended: this model will comply with requests the base model would refuse, including some it shouldn't. There is no safety filtering layered on top. You are responsible for how you deploy it — don't put this behind an unmoderated public-facing endpoint serving third parties. It inherits MiniCPM5-2B's factual limitations and biases; abliteration removes refusal directions, it doesn't add capability or judgment.

License

Inherits the Apache 2.0 license from the base model.

Related

• MiniCPM5-1B-heretic
• RACER IS OP — Heretic Models — full collection
• RaceBench-MiniCPM5-heretic
• Qwen3-1.7B-heretic

---

Base model: openbmb/MiniCPM5-2B

Original MiniCPM5-2B model card (click to expand)

See the base model card at openbmb/MiniCPM5-2B for the original architecture, training details, requirements, and citation.

Abliteration hardware

Recorded in reproduce/reproduce.json:

• GPU: Tesla T4 (14.56 GB VRAM), Tesla T4 (14.56 GB VRAM)
• CPU: Intel(R) Xeon(R) CPU @ 2.00GHz
• OS: Linux-6.12.90+-x86_64-with-glibc2.35
• CUDA / driver: CUDA 13.0, driver 580.159.04
• Software: PyTorch 2.14.0+cu130, Heretic 1.4.0

Heretic notes that KL-divergence and refusal numbers can vary slightly with hardware and library versions.